MCP servers that deploy & manage cloud infrastructure
Manage deployments, containers, and resources on AWS, Vercel, K8s & more.
28 servers · Last updated August 1, 2026
TL;DR: These servers let an agent operate infrastructure — deploying apps, managing containers and clusters, and querying cloud resources on AWS, Vercel, Netlify, Railway, Kubernetes, Docker and Terraform. Powerful and dangerous: scope credentials tightly and prefer read/plan operations over unguarded applies.
Bottom line: if you only try one, kubefwd is the most popular option for this (4,125★). 27 more compared below.
Compare 28 servers
| Server | Transport | Auth | Verified | Stars | Tools for this |
|---|---|---|---|---|---|
| kubefwd | Local (stdio) | No auth | — | 4,125 | list_k8s_services, add_service, remove_service +8 |
| Radar | Remote (HTTP) | No auth | — | 2,425 | top_resources, list_resources, get_resource +4 |
| Grafana MCP Server | Local (stdio) | API key | 2,100 | get_panel_image | |
| Kubernetes MCP Server | Local (stdio) | No auth | — | 1,700 | pods_list, pods_list_in_namespace, pods_get +12 |
| Kubernetes MCP Server | Local (stdio) | No auth | 1,400 | kubectl_create / kubectl_apply / kubectl_delete, cleanup_pods, explain_resource / list_api_resources / ping | |
| Terraform MCP Server | Local (stdio) | No auth | 1,400 | list_stacks / get_stack_details | |
| kubectl-mcp-server | Local (stdio) | No auth | — | 913 | get_pods, diagnose_pod_crash, get_deployments +6 |
| K8M | Remote (SSE) | API key | — | 849 | list_clusters, scale_deployment, restart_deployment +34 |
| Docker MCP Server | Local (stdio) | No auth | 720 | list_containers, create_container, run_container +10 | |
| Cloud Run MCP Server | Local (stdio) | OAuth | — | 617 | deploy-file-contents, list-services, get-service +2 |
| Railway MCP Server | Local (stdio) | API key | 500 | service-list, service-create-from-repo, service-create-from-image +5 | |
| EdgeOne Pages MCP | Local (stdio) | No auth | — | 420 | deploy_html, deploy_folder, deploy_folder_or_zip |
| MCP K8S Go | Local (stdio) | No auth | — | 382 | List, get, create and modify Kubernetes resources, List Kubernetes pods, Get Kubernetes pod logs +1 |
| tfmcp | Local (stdio) | No auth | — | 369 | apply_terraform, list_terraform_resources, get_resource_dependency_graph |
| Netlify MCP Server | Local (stdio) | API key | 320 | netlify-user-services, netlify-project-services, netlify-deploy-services +1 | |
| MCP SSH Manager | Local (stdio) | No auth | — | 278 | ssh_service_status, ssh_deploy |
| Kom (Kubernetes Operations Manager) | Local (stdio) | No auth | — | 148 | list_k8s_clusters, scale_k8s_deployment, restart_k8s_deployment +40 |
| mcp-k8s | Local (stdio) | No auth | — | 146 | get_api_resources, get_resource, list_resources +3 |
| Alibaba Cloud Ops MCP Server | Local (stdio) | API key | — | 119 | StartInstances, StopInstances, RebootInstances +13 |
| DigitalOcean MCP | Local (stdio) | API key | — | 80 | list_deployments, create_deployment, get_deployment +3 |
| Heroku MCP Server | Local (stdio) | API key | — | 78 | deploy_to_heroku, deploy_one_off_dyno |
| Bitrise MCP Server | Remote (HTTP) | OAuth | — | 39 | list_available_stacks, codepush_list_deployments, codepush_get_deployment +5 |
| EntraID MCP Server (Microsoft Graph) | Local (stdio) | API key | — | 39 | list_service_principals, get_service_principal_by_id, create_service_principal +2 |
| APISIX MCP Server | Local (stdio) | API key | — | 38 | get_resource, delete_resource, create_service / update_service / delete_service |
| MikroMCP | Local (stdio) | No auth | — | 35 | apply_plan |
| mcp-k8s-eye | Local (stdio) | No auth | — | 28 | resource_get, resource_list, resource_create_or_update +9 |
| Vercel MCP Server (remote) | Remote (HTTP) | OAuth | 0 | list_deployments / get_deployment, get_deployment_build_logs, deploy_to_vercel | |
| Pulumi MCP Server | Local (stdio) | API key | — | 0 | pulumi-registry-list-resources, pulumi-registry-get-resource, pulumi-cli-stack-output +1 |
The servers
Bulk Kubernetes port forwarding with an MCP server that lets AI assistants forward services, inspect pods, and monitor traffic.
Built-in MCP server for the open-source Kubernetes UI — lets AI assistants query and operate your cluster.
Grafana's official server for dashboards, Prometheus/Loki queries, alerts, and incidents.
Go-native MCP server for Kubernetes and OpenShift — manage pods, resources, Helm, Tekton, and more.
Popular community server wrapping kubectl and Helm for cluster management via natural language.
HashiCorp's official server for Terraform Registry providers, modules, and IaC workflows.
Control your entire Kubernetes infrastructure through natural language conversations with AI.
AI-driven mini Kubernetes dashboard with a built-in MCP server exposing 49 cluster operation tools
Community server to manage Docker containers, images, volumes, and Compose stacks via agents.
Deploy apps and folders to Google Cloud Run directly from MCP-compatible AI agents.
Community server to manage Railway.app projects, services, and deployments via your agent.
Deploy HTML, folders, or full-stack projects to Tencent EdgeOne Pages and get a public URL.
Golang-based MCP server connecting to Kubernetes — list, get, create and modify resources, read logs, run commands.
Terraform MCP server letting AI assistants read, plan, apply, analyze, and manage Terraform infrastructure.
Netlify's official server to create, build, deploy, and manage Netlify sites from your agent.
Manage multiple SSH servers from Claude Code or OpenAI Codex — execute commands, transfer files, backups, DB ops, health monitoring.
Multi-cluster Kubernetes MCP server with 59 tools for resources, pods, nodes, deployments and YAML.
Kubernetes MCP server: manage cluster resources and Helm releases through MCP tools, with fine-grained write controls.
Operate Alibaba Cloud resources (ECS, VPC, RDS, OSS, CloudMonitor, OOS) from AI assistants via MCP.
DigitalOcean's App Platform MCP: deploy apps, tail logs, roll back. Archived — superseded by mcp-digitalocean.
Manage Heroku apps, dynos, add-ons, pipelines, and Postgres from your AI agent via the official Heroku CLI.
Manage Bitrise apps, builds, artifacts, pipelines and releases from your AI assistant.
Manage Microsoft Entra ID via Graph API — users, groups, apps, MFA, sign-in logs, conditional access and more.
Manage Apache APISIX gateway resources through natural language via the Admin API.
Production-grade MCP server for MikroTik RouterOS — 117 typed, auditable tools for safe AI-native network automation.
Manage Kubernetes clusters and diagnose workload health via MCP.
Vercel's hosted OAuth server for inspecting projects, deployments, and logs from your AI client.
Give agents Pulumi powers: browse the provider registry, preview and deploy stacks, and search deployed resources.
Use these in a stack
FAQ
Is it safe to give an AI cloud-infra access?
Only with least-privilege credentials and, ideally, plan/read-only modes. An agent with broad cloud write access can do real damage — gate destructive actions behind human approval.
Which platforms have infra MCP servers?
AWS, Vercel, Netlify, Railway, Kubernetes, Docker and Terraform all have servers here — pick the ones matching your deployment targets.